INDEX

Explanations

phrases or words related to information, frequently emphasizing intensity or urgency, such as 'infusing', 'infidelity', or 'infanticide'

oai_token-act-pair · gpt-3.5-turbo

instances of the prefix "inf" which indicate a focus on information-related topics

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Blitz

-0.76

å§«

-0.76

BOOK

-0.73

BY

-0.66

OPLE

-0.66

 Boll

-0.65

Gallery

-0.64

Bil

-0.64

 Athletic

-0.63

 Mata

-0.62

POSITIVE LOGITS

requent

1.22

usions

1.20

ighting

1.17

requently

1.16

ractions

1.16

idelity

1.16

luence

1.15

erno

1.12

raction

1.10

ruct

1.09

Activations Density 0.016%

phrases or words related to information, frequently emphasizing intensity or urgency, such as 'infusing', 'infidelity', or 'infanticide'

instances of the prefix "inf" which indicate a focus on information-related topics

No Comments

No Known Activations